kvmppc: support guest shutdown - guest part

PATCH In review [email protected] 2 participants 4 messages Apr 7, 2008

Subsystems (2)

PowerPC KVM PowerPC

lore

Share & tools

lore

[email protected]

author

2008-04-07 12:37 UTC #0 lore raw

PATCH 1/1 kvmppc: support guest shutdown - guest part

From: Christian Ehrhardt <[email protected]>

This is the guest portion of this patch, it submits a trap with the special immediate value where the native code would enter a infinite loop with irq's disabled (no native shutdown available).
This is only a interim solution until the hypercall abi/api is fully defined. Then we will recode it in according to that specification. I would wait for these final hypercall variant before we go to [email protected] with these patch, but for now this patch could stay in our patch queues because it easens daily usage e.g. scripted testing. Note - since qemu defaults to reboot the guest on shutdown you might need to the -no-reboot command line option to qemu to get what is expected.

Signed-off-by Christian Ehrhardt <[email protected]>

[diffstat]

 kernel/setup-common.c |    4 ++++
 kvm/Kconfig           |   11 ++++++++++-
 2 files changed, 14 insertions(+), 1 deletion(-)

---

arch/powerpc/kernel/setup-common.c +4 -0

@@ -146,6 +146,10 @@ void machine_halt(void)
 #endif
 	printk(KERN_EMERG "System Halted, OK to turn off power\n");
 	local_irq_disable();
+#ifdef CONFIG_KVM_PPC_GUEST
+	/* FIXME port to the proper hypercall API/ABI when defined */
+	__asm__ __volatile__("twi 31,0,1");
+#endif
 	while (1) ;
 }

arch/powerpc/kvm/Kconfig +10 -1

@@ -35,7 +35,7 @@ config KVM_POWERPC
 	bool
 
 config KVM_POWERPC_440
-	tristate "KVM guest support for PowerPC 440"
+	tristate "KVM host support for PowerPC 440 guests"
 	depends on KVM && 44x
 	select KVM_POWERPC
 	---help---

@@ -52,5 +52,14 @@ config KVM_PPC_VIRTIO
 # the virtualization menu.
 source drivers/virtio/Kconfig
 
+
+config KVM_PPC_GUEST
+	bool "Paravirt guest extensions for KVM on POWERPC"
+	depends on 44x
+	---help---
+	  This add support of paravirt features for KVM on POWERPC
+
+
+
 endif # VIRTUALIZATION

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Register now and save $200. Hurry, offer ends at 11:59 p.m., Monday, April 7! Use priority code J8TLD2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
kvm-ppc-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-ppc-devel

[email protected]

author

2008-04-07 12:37 UTC #1 lore raw

PATCH 1/1 kvmppc: support guest shutdown - host part

From: Christian Ehrhardt <[email protected]>

This is the host portion of this patch, it detects a trap with the special immediate value and then exits with the kvm exit_reason for shutdown. This is only a interim solution until the hypercall abi/api is fully defined. Then we will recode it in according to that specification. I would wait for these final hypercall variant before we go to [email protected] with these patch, but for now this patch could stay in our patch queues because it easens daily usage e.g. scripted testing.

Signed-off-by Christian Ehrhardt <[email protected]>

[diffstat]

 arch/powerpc/kvm/emulate.c    |    5 +++++
 arch/powerpc/kvm/powerpc.c    |    4 ++++
 include/asm-powerpc/kvm_ppc.h |    1 +
 3 files changed, 10 insertions(+)

---

arch/powerpc/kvm/emulate.c +5 -0

@@ -229,6 +229,11 @@ int kvmppc_emulate_instruction(struct kv
 
 	switch (get_op(inst)) {
 	case 3:                                                 /* trap */
+		if (get_d(inst) = 1) {
+			/* FIXME port to final hypercall API when defined */
+			printk(KERN_INFO"Guest requested shutdown\n");
+			return EMULATE_SHUTDOWN;
+		}
 		printk("trap!\n");
 		kvmppc_queue_exception(vcpu, BOOKE_INTERRUPT_PROGRAM);
 		advance = 0;

arch/powerpc/kvm/powerpc.c +4 -0

@@ -354,6 +354,10 @@ int kvmppc_handle_exit(struct kvm_run *r
 			 * report it to userspace. */
 			run->hw.hardware_exit_reason = ~0ULL << 32;
 			run->hw.hardware_exit_reason |= vcpu->arch.last_inst;
+			r = RESUME_HOST;
+			break;
+		case EMULATE_SHUTDOWN:
+			run->exit_reason = KVM_EXIT_SHUTDOWN;
 			r = RESUME_HOST;
 			break;
 		default:

include/asm-powerpc/kvm_ppc.h +1 -0

@@ -39,6 +39,7 @@ enum emulation_result {
 	EMULATE_DO_MMIO,      /* kvm_run filled with MMIO request */
 	EMULATE_DO_DCR,       /* kvm_run filled with DCR request */
 	EMULATE_FAIL,         /* can't emulate this instruction */
+	EMULATE_SHUTDOWN,     /* shutdown guest requested (no kvm_run data) */
 };
 
 extern const unsigned char exception_priority[];

Jimi Xenidis

2008-04-07 14:13 UTC #2 lore raw

Re: [kvm-ppc-devel] [PATCH] kvmppc: support guest shutdown - guest

In reply to [email protected]

Quoted 30 lines · On Apr 7, 2008, at 8:37 AM, [email protected] wrote:

From: Christian Ehrhardt <[email protected]>
This is the guest portion of this patch, it submits a trap with the special
immediate value where the native code would enter a infinite loop with irq's
disabled (no native shutdown available).
This is only a interim solution until the hypercall abi/api is fully defined. Then we will recode it in according to that specification. I would wait for these final hypercall variant before we go to [email protected] with these patch, but for now this patch could stay in
our patch queues because it easens daily usage e.g. scripted testing. Note - since qemu defaults to reboot the guest on shutdown you might need to
the -no-reboot command line option to qemu to get what is expected.
Signed-off-by Christian Ehrhardt <[email protected]>
[diffstat]
 kernel/setup-common.c |    4 ++++
 kvm/Kconfig           |   11 ++++++++++-
 2 files changed, 14 insertions(+), 1 deletion(-)
---
arch/powerpc/ renamed from arch/powerpc/kernel/setup-common.c +0 -0
kernel/setup-common.c

arch/powerpc/kernel/setup-common.c void machine_halt(void) · line 148

Quoted 3 lines

arch/powerpc/kernel/setup-common.c +0 -0

@@ -146,6 +146,10 @@ void machine_halt(void)
 #endif
 	printk(KERN_EMERG "System Halted, OK to turn off power\n");
 	local_irq_disable();

hmm.. shouldn't you just be defining ppc_md.halt in your setup so it can be called in this function above this line?

Functionally I would assume this "zombies" the guest rather then completely destroys it?
This is important if I want to externally debug said state.

You may also want to consider defining ppc_md.power_off as well which would automatically and completely destroy the guest.

arch/powerpc/kernel/setup-common.c void machine_halt(void) · line 152

Quoted 4 lines

arch/powerpc/kernel/setup-common.c +4 -0

+#ifdef CONFIG_KVM_PPC_GUEST
+	/* FIXME port to the proper hypercall API/ABI when defined */
+	__asm__ __volatile__("twi 31,0,1");
+#endif

Have you considered using mtmsr(MSR_WE) which puts you in wait with no hope of coming back, unless forced by probe.

Quoted 34 lines

arch/powerpc/kernel/setup-common.c +0 -0

 	while (1) ;
 }

arch/powerpc/kvm/Kconfig +10 -3

@@ -35,7 +35,7 @@ config KVM_POWERPC
 	bool
 
 config KVM_POWERPC_440
-	tristate "KVM guest support for PowerPC 440"
+	tristate "KVM host support for PowerPC 440 guests"
 	depends on KVM && 44x
 	select KVM_POWERPC
 	---help---

@@ -52,5 +52,14 @@ config KVM_PPC_VIRTIO
 # the virtualization menu.
 source drivers/virtio/Kconfig
 
+
+config KVM_PPC_GUEST
+	bool "Paravirt guest extensions for KVM on POWERPC"
+	depends on 44x
+	---help---
+	  This add support of paravirt features for KVM on POWERPC
+
+
+
 endif # VIRTUALIZATION
 
 
---------------------------------------------------------------------- 
---

This SF.net email is sponsored by the 2008 JavaOne(SM) Conference

Quoted 8 lines

Register now and save $200. Hurry, offer ends at 11:59 p.m., Monday, April 7! Use priority code J8TLD2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://
java.sun.com/javaone
_______________________________________________
kvm-ppc-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-ppc-devel

Christian Ehrhardt

author

2008-04-07 15:23 UTC #3 lore raw

Re: [kvm-ppc-devel] [PATCH] kvmppc: support guest shutdown - guest

In reply to [email protected]

Quoted 27 lines · Jimi Xenidis

Quoted 25 lines · On Apr 7, 2008, at 8:37 AM, [email protected] wrote:
From: Christian Ehrhardt <[email protected]>
This is the guest portion of this patch, it submits a trap with the special
immediate value where the native code would enter a infinite loop with irq's
disabled (no native shutdown available).
This is only a interim solution until the hypercall abi/api is fully defined. Then we will recode it in according to that specification. I would wait for these final hypercall variant before we go to [email protected] with these patch, but for now this patch could stay in
our patch queues because it easens daily usage e.g. scripted testing. Note - since qemu defaults to reboot the guest on shutdown you might need to
the -no-reboot command line option to qemu to get what is expected.
Signed-off-by Christian Ehrhardt <[email protected]>
[diffstat]
 kernel/setup-common.c |    4 ++++
 kvm/Kconfig           |   11 ++++++++++-
 2 files changed, 14 insertions(+), 1 deletion(-)
---

Quoted 11 lines

arch/powerpc/kernel/setup-common.c void machine_halt(void) · line 148
Quoted 8 lines
arch/powerpc/kernel/setup-common.c +0 -0
@@ -146,6 +146,10 @@ void machine_halt(void)
 #endif
     printk(KERN_EMERG "System Halted, OK to turn off power\n");
     local_irq_disable();
hmm.. shouldn't you just be defining ppc_md.halt in your setup so it can be called in this function above this line?

I'll look into that - thx Jimi

Quoted 3 lines

Functionally I would assume this "zombies" the guest rather then completely destroys it?
This is important if I want to externally debug said state.

hmmm ... I started with the demand to get it destroyed on halt.

Quoted 10 lines

You may also want to consider defining ppc_md.power_off as well which would automatically and completely destroy the guest.
arch/powerpc/kernel/setup-common.c void machine_halt(void) · line 152
Quoted 4 lines
arch/powerpc/kernel/setup-common.c +4 -0
+#ifdef CONFIG_KVM_PPC_GUEST
+    /* FIXME port to the proper hypercall API/ABI when defined */
+    __asm__ __volatile__("twi 31,0,1");
+#endif
Have you considered using mtmsr(MSR_WE) which puts you in wait with no hope of coming back, unless forced by probe.

atm I want it dead ;-).
But you'r right - eventually we should implement the differences between halt, poweroff, ... in the final version

Quoted 42 lines

Quoted 41 lines
arch/powerpc/kernel/setup-common.c +0 -0
     while (1) ;
 }
 
arch/powerpc/kvm/Kconfig +10 -2
@@ -35,7 +35,7 @@ config KVM_POWERPC
     bool
 
 config KVM_POWERPC_440
-    tristate "KVM guest support for PowerPC 440"
+    tristate "KVM host support for PowerPC 440 guests"
     depends on KVM && 44x
     select KVM_POWERPC
     ---help---
@@ -52,5 +52,14 @@ config KVM_PPC_VIRTIO
 # the virtualization menu.
 source drivers/virtio/Kconfig
 
+
+config KVM_PPC_GUEST
+    bool "Paravirt guest extensions for KVM on POWERPC"
+    depends on 44x
+    ---help---
+      This add support of paravirt features for KVM on POWERPC
+
+
+
 endif # VIRTUALIZATION
 
 
-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Register now and save $200. Hurry, offer ends at 11:59 p.m., Monday, April 7! Use priority code J8TLD2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
kvm-ppc-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-ppc-devel

Signature · 13 lines

Grüsse / regards, 
Christian Ehrhardt
IBM Linux Technology Center, Open Virtualization

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Register now and save $200. Hurry, offer ends at 11:59 p.m., 
Monday, April 7! Use priority code J8TLD2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
kvm-ppc-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-ppc-devel